TEORÍAS CIENTÍFICAS DEL LIDERAZGO PEDAGÓGICO

The previous section described the history, controversy, and potential utility of measuring audience perceptions of message effectiveness as a means of selecting persuasive messages about e-cigarettes that are likely to be effective. PME is often situated within theories of reasoned action or planned behavior within an attitude-intention-behavior model (Ajzen, 1991; Fishbein & Ajzen, 2011). These theories that have guided the design and implementation of PME have sought to examine both how perceptions of an ad’s attributes and likely impact on individuals influence subsequent attitudes, intentions and behaviors (see Figure 2) (Davis, Nonnemaker, Farrelly, & Niederdeppe, 2010; Mitchell & Olson, 1981; Shimp, 1981) as well as how previously held attitudes affect processing of persuasive messages (Dillard & Ye, 2008; Hullett & Boster, 2001).

As stated above, there is room for growth in the literature in designing PME measures that more directly correspond to the outcomes of interest and the specific aims of advertisements being assessed. Likewise, there is room for growth in operationalizing two key dimensions of theories of reasoned action behavioral models in assessing the effects of anti-vaping

advertisements. The rest of this section will deal with the first dimension: designing assessments of baseline attitudes that are more in correspondence with (and likely to be affected by) the

intended purpose of anti-tobacco advertisements. The following section will describe the second dimension: leveraging theories and methods from egocentric (personal network) network

research to more fully understand the role an individual’s social environment plays in determining baseline attitudes and subsequent reactions to anti-vaping messages.

Likewise, there is a gap in the current literature for examining post-advertisement exposure outcomes beyond quit intentions or smoking behaviors (see Bigsby et al., 2013; Davis et al., 2013; Davis et al., 2017). Measurements that allow for more detailed assessments of attitude change as outcomes of interest may provide empirically and theoretically valuable insights into the thresholds needed for a persuasive message to cascade from altering the different steps of the attitudes-intentions-behaviors models of behavior change. Previous research has examined how individuals’ expectations of the outcomes (or outcome expectancies [OE]) associated with particular behaviors influence their adoption and sustained implementation of those behaviors provides. Insights from these literatures may help bridge the aforementioned gaps in PME assessments. The purpose of this section is to provide a working definition of outcome expectancies, situate them within relevant literatures, demonstrate how they have been incorporated into previous tobacco behavior literatures, and explain why outcome expectancies should be utilized as baseline examinations of relevant attitudes in assessments of anti-e-cigarette advertisements.

Conceptual definition and theoretical tradition

From a conceptual standpoint, determining the expected outcomes associated with a behavior in order to determine underlying causes or beliefs about that particular behavior has a logical elegance. Jones and colleagues (2001), in a review of expectancy theory and its relation to alcohol dependence research, invoke the concept of Occam’s razor to describe how a single

measurement of outcome expectancies can assess a construct that includes multiple dimensions. This theoretical assertion, that outcome expectancies can be comprised of competing dimensions, is similar to one of the key theoretical arguments surrounding PME. As stated in the previous section, PME can assess perceptions of a message as well as its likely effects. OE scales, on the other hand, can assess a respondents’ expectations about multiple dimensions associated with vaping such as social (“will I be ostracized?), health (“will vaping harm my lungs?), or personal experience (“will vaping help alleviate stress?”)outcomes (Barker et al., 2018).

This study adapted the definition promoted by Jones et al. (2001) and consider outcome expectancies as structures in long-term memory that impact cognitive processes governing current and future behaviors associated with those structures (p. 59). The term outcome

expectancy is closely related and often used in tandem with the concept of outcome expectancies (Bandura, 1986) as both concepts are theorized to mediate behavior through the assessment of anticipated outcomes associated with that behavior (Bandura, 1986; Jones et al., 2001) The concept of an outcome expectancy can, thus, be illustrated by the answer to the question: “Well, what did you think was going to happen?”

Outcome expectancies have their empirical and theoretical roots in social learning

theories Bandura (1986; 2001). Bandura (1977; 1986) posits that the source of individual beliefs about outcomes can be traced from three main sources. The first source is symbolic thinking, or the imagined consequences that an individual believes might arise if he or she should perform a particular action (Bandura, 1977; Fouad & Guillen, 2006). An individual who decides to stay in and write a dissertation rather than going out for the evening with his or her friends has used symbolic thinking if the expected consequences of either option influence his or her behavioral

decision. Individuals can also model their behavior through vicarious experiences or models demonstrating positive or negative outcomes of a behavior.

Research into the effects of e-cigarette advertising on adolescent risk perceptions is a good example of the role vicarious observation can have on developing outcome expectancies. Results from recent studies have demonstrated exposure to e-cigarette advertising predicts more positive OE among adolescents, regardless of previous use (Phua et al., 2017; Pu & Zhang, 2017). Recent research examining the effects of pro-vaping messages on young adults suggests exposure to vaping advertisements as well as user-created social media groups promoting vaping can have negative outcomes on outcome expectancies of quitting and self-efficacy to stop using e-cigarettes (Phua, 2018).

Finally, Bandura (1977) posits that OEs can form from incentive values of an outcome or a consequence of the action. Fouad and Guillen (2006) describe how effort put into careers can be altered by environmental incentives such as compensation or perceptions of social support. Although Bandura’s conceptualization of outcome expectancies includes both social and

behavioral impacts associated with a behavior, other theoretical traditions contend with the likely differential impact social and behavioral outcomes may hold. Within theories of reasoned action, an individual’s social support or social influences are understood as impacting underlying salient normative beliefs, while OE are examples of salient behavioral beliefs (Armitage & Christian, 2003). Thus, the social support structure or social environment in which a person is enmeshed can be understood as a causal determinant of normative beliefs about a behavior (e.g., how socially acceptable using e-cigarettes is) while outcome expectancies are representative of behavioral beliefs (e.g., what will happen to my mood, to my health, or to my social standing if I use this product). These two dimensions are understood in theories of reasoned action to

additively impact the likelihood that an individual will either choose or not choose to use an addictive substance or engage in a behavior.

Bandura’s development of OE is rooted in psychological theories and models that can be traced back to Tolman’s (1932) cognitive construction of expectancy. Tolman (1932;

paraphrased in Fouad & Guillen, 2006) defined the cognitive aspect of learning as a mediating variable derived from animals learning about what would happen if they performed a particular action (Fouad & Guillen, 2006, p. 132). Tolman’s (1932) concept of expectancy, or “purposive behaviorism” was an indicator of a paradigm shift from behaviorist psychological models to cognitive models. Fouad and Guillen (2006) interpret Tolman’s learning theory as reliant on expected rewards or punishments as integral elements in facilitating learning. For example, a rat that runs through a maze over and over again, getting faster each time, is interpreted as learning the turns of the maze in tandem with a growing expectation of cheese as an outcome of finishing the puzzle.

As cognitive models of psychology overtook behaviorist interpretations, the role of OE in learning behaviors was investigated more. Stacy and colleagues (1990) describe the role that Bolles’s (1972) expectancy theory played in shaping understanding of OE. Building off of Tolman’s (1932) work, Bolles (1972) reviewed behaviors demonstrated by animals in clinical trials in order to arrive at his definition of an expectancy as information “about a new order of things in the environment” (p. 402). In much the same way that Tolman’s (1932) hypothesis about the role of expectancies came during a transitionary period between paradigms, Bolles’s (1972) expectancy theory sought to redirect the popular motivation/reinforcement theory in which individuals’ behaviors were theorized to be a product of motivation altered by direct reinforcements received as a result of that behavior. For example, a student that misses school

and is punished by her principal may be less likely to purposefully miss again. Expectancy theory (Bolles, 1972) maintains many of the basic moving parts of motivation/reinforcement theory but allows for the role of outcomes that have not been experienced directly to influence behavior. To return to the truancy example above, expectancy theory would allow for the public recrimination of a student who has skipped class to influence the decision-making processes of students who see the punishment.

Jones and colleagues (2001) argue that the inclusion of indirectly observed outcomes to influence subsequent behavior allowed for the potential for outcomes that are illogically formed or misinformed to influence subsequent behaviors so long as they are held and believed by the individual under analysis. This relaxation of the etiological restraints for allowing expected results to influence motivations and, ultimately, behaviors allowed for a wider range of social circumstances to exhibit influences on outcomes of interest. As Jones and colleagues (2001) describe, this flexibility to examine both logical and illogical expected outcomes as antecedents of behaviors was a natural fit for examining the processes that instigated and supported

alcoholism in the addiction literatures of the 1980s. As more scholars began incorporating expectancy theory or social learning theories into a wide array of literatures, scholars started to examine subgroups of OE and how those subgroups might correlate with particular types of behavior.

Stimulus and response expectancies

One of the most influential scholars examining OE subgroups was Kirsch (1997) whose response expectancy theory argued for a bifurcation of OE into stimulus and response

expectancies. A stimulus expectancy, as Kirsch (1997) describes, is the type of expectancy that is most often examined in theories like TPB (Ajzen, 1991) or SLT (Bandura, 1977, 1986). This

type of expectancy can be likened to an adolescent’s expectancy that using an e-cigarette will make her more ”cool” within her social spheres. As a result of this expectancy, the adolescent might spend more time with friends who also use e-cigarettes or spend more time with friends outside of school where she can use e-cigarettes more freely. The stimulus expectancy had an effect on behavior in this case—i.e. the student uses e-cigarettes and spends more time with friends outside of school—and may have an indirect effect on the behavior’s outcome—feeling more included in a social circle. Stimulus expectancies are thus expectancies about outcomes that are not fully under the individual’s control (Kirsch, 1997, p. 69). In other words, the stimulus expectancy that using e-cigarettes will lead to being more popular is mediated through external factors such as the adolescent’s social circle’s involvement and perceptions that are outside the immediate outcomes associated with the stimulating behavior (using an e-cigarette).

Kirsch (1997) distinguishes this type of expectancy from response expectancy, which he argues is beneficial in understanding the types of OE that can affect the ability of an individual to enter hypnosis, ascribe benefits to placebos, or seek stimulation from addictive substances (p. 70). Response expectancy is distinguishable from stimulus expectancy because the expectations are either directly confirmed or dismissed as a result of the behavior to which that outcome expectation is tied.

For example, if an individual believes that smoking an e-cigarette will give them a

pleasant sensation or that drinking will make them feel more at ease, the outcome is an automatic response to the behavior that individual engages in. Once again, in keeping with expectancy theory, neither of these types of expectations need to hold logical or even realistic grounds (Jones et al., 2001). The student who believes using an e-cigarette will make her more popular

spends more time away from formal institutions (school) with her friends and not directly because of using e-cigarettes. Likewise, Kirsch (1997) describes empirical studies in which response expectancies are studied have demonstrated that individuals who drink caffeinated coffee or non-alcoholic beer will report “feeling” the effects of those drugs even if they contain none of the chemicals that would cause those feelings to occur.

Within the context of addictive substances literatures, OE are commonly interpreted as mediating variables within risk assessments that can significantly alter the susceptibility or usage behaviors for a variety of addictive substances (Sitkin & Pablo, 1992). This conceptualization of outcome expectancies as a mediating factor is theorized to exhibit greater influence in situations in which the “true” outcome of an action or behavior is unknown or ambiguous (Sitkin & Pablo, 1992). While not often explicated in the literature, contemporary studies often incorporate measures of both stimulus and response OE in decisional models for starting to use and continuing use of e-cigarettes. Kirsch’s (1977) bifurcation of stimulus and response OE is indicative of the malleability of the concept in examining determinants of health outcomes. This malleability and ability to incorporate OE measures into theories of reasoned action as well as social learning theories were important factors in the proliferation of OE measures throughout studies of alcohol consumption in the 1980s and 1990s (see Figure 3). As previously stated above, the theories underlying PME place a priority on valid assessment of attitudes related both to products and behaviors as well as the persuasive messages either promoting or admonishing them. The theoretical examination of OE will conclude with a brief demonstration of how OE has been used as a measure of existing attitudes toward addictive substances and how this usage might better inform PME research.

Figure 3: Reasoned action model including outcome expectancies

4

OE and attitudes

Previous public health research into alcohol consumption in the 1980s and 1990s examined the extent to which OE were associated with existing attitudes toward drinking. The conceptual definition for OE adopted in this study: structures in long-term memory that impact cognitive processes governing current and future behaviors associated with those structures (adapted from Jones et al., 2001, p. 59) is reminiscent of descriptions of attitudes in commonly used theories of reasoned action that have been used to situate attitudes toward an advertisement within the attitude-intention-behavior model of behavioral change (Ajzen, 1991; Fishbein, 1979). Kuther (2002) asserts that both the theory of reasoned action (Fishbein & Ajzen, 1975) and theory of planned behavior (Ajzen, 1991) conceptualize attitudes as beliefs or expectations about behavioral outcomes and evaluations of behavioral outcomes.

Goldman and colleagues (1991) argue that attitude and expectancy could be examinations of the same unidimensional construct that may “merely reflect differing points of emphasis in various social/cognitive models of behavior” (p. 143). This theoretical assertion is supported by Leigh (1989), who argues predicting drinking behavior can be done “as easily and as reliably with attitudes as with expectancies” (p. 366). One of the drawbacks of incorporating OE into examinations of alcohol use at the time, the author acknowledges was that attitude research was backed by more research and theory (Leigh, 1989). This assertion has been supported by empirical data from Stacy and colleagues (1990) who demonstrated greater explanatory power for both alcohol-use intentions and subsequent behaviors by assessing relevant positive and negative outcome expectancies rather than traditional rational decision theories measures of attitudes.

The potential differential explanatory power of outcome expectancies as opposed to traditional measures of attitudes is described in detail by Kuther (2002). The author details a number of studies in which alcohol expectancies have outperformed traditional measures of attitude (see p. 39), asserting that the relative lack of specificity in the construction of a number of attitude scales informed by the theory of planned behavior compared to expectancy scales may be a reason for the differential explanatory power. Kuther (2002) accepts the theoretical

similarities between outcome expectancies and attitude, but asserts that measurement differences, i.e. expectancy measurements of specific outcomes (e.g. feeling relaxed) versus Ajzen-style generalized outcome measurements (e.g. feeling pleasant/unpleasant) (p. 40).

The examination of specific outcomes associated with addictive behaviors also allows researchers an opportunity to examine respondent evaluations of salient outcomes. Stacy and colleagues (1990) describe how OE can be constructed to not only examine the likelihood of two specific outcomes (e.g. that smoking an e-cigarette will taste good; lead to addiction), these specific outcomes can also be valued by evaluations of likelihood or perceived severity. As PME is theoretically and empirically linked with how existing attitudes about behaviors or products influence perceptions of persuasive messages about those behaviors or products, OE measurements that allow for specific examinations of likely outcomes as well as their relative valuation should increase correspondence between antecedent beliefs and their subsequent changes following the introduction of a message stimulus. The final part of this section will examine selected previous research that has studied the role OE play in e-cigarette usage and explain how the current study can build off of these results.

OE and e-cigarette usage

Much like the variety of nomenclatures used for PME described in the previous section, researchers examining OE in the context of addictive substances have used a number of different titles for their measurements. In addictive substance literatures, OE have been studied within the context of risk perceptions (e.g. Agaku et al., 2018; Lippert, 2016), outcome expectations

(Barnett, Lorenzo, & Soule, 2017; Wilkinson et al., 2009), and outcome expectancies (Pokhrel et al., 2014; Southwick, Steele, Marlatt, & Lindell, 1981; Stacy, Dent, et al., 1990). Information about how outcome expectancies affect adolescent and young adult usage of e-cigarettes has been gathered using a variety of methods. Qualitative interviews (Pokhrel et al., 2014), surveys (Harrell et al., 2015) and focus groups (Wagoner et al., 2016) have all been used to examine how young adults and adolescents conceptualize OE associated with e-cigarette use. Although the methods of data collection are varied, there is a historical context for using multiple methods to investigate outcome expectancies in relation to an addictive behavior. Jones and colleagues (2001) describe the wide variety of methods used to derive outcome expectancies measures in the 1980s and 1990s to study alcoholism. Studies published within that time frame included many of the data collection methods described above in an effort to cast the widest net possible to determine which OE were most correlated and most predictive of alcohol consumption behaviors.

Health outcome expectancies

Results from initial research into e-cigarettes have suggested both social and health-related outcomes may be important in mediating usage behaviors. Perhaps the broadest finding that can be extrapolated to a large number of studies is the consistent finding that individuals perceive e- cigarettes to be less directly harmful to an individual’s health than traditional cigarettes. This

finding has been demonstrated in adolescents (Amrock, Lee, & Weitzman, 2016), young adults (Pokhrel, Lam, Pagano, Kawamoto, & Herzog, 2018; Pokhrel et al., 2014) and even hospitalized smokers (Hendricks et al., 2015). Previous research has suggested that OE about the relative lack of harm associated with e-cigarettes compared to traditional cigarettes could be linked to misinformed beliefs about a lack of nicotine or harmful substances in e-cigarettes (Wagoner et